基于随机化分析的等价空间故障检测方法

A randomized approach to parity space-based fault detection

基于等价空间的故障检测可实现残差对控制输入与未知状态的解耦,已成为提高系统安全性与可靠性的重要技术手段。针对难以同时降低故障漏报率与误报率的问题,提出一种随机化分析辅助的等价空间故障检测方法。首先,在未知输入概率分布未知的情况下,设计基于等价空间的残差产生器,并应用Chernoff界引理,给出故障误报率的随机化分析方法。其次,针对零故障误报率阈值选取的强保守性问题,给出一定故障误报率容许条件下的阈值选取方法,实现了故障误报率与可检测率的均衡设计。最后以某型无人机飞行控制系统进行仿真实验,验证了提出方法的有效性。

Parity space-based fault detection can realize residual’s decoupling between control inputs and unknown system states and has become an important technique to improve system safety and reliability. To resolve the difficulty of reducing both the false alarm rate and the missing alarm rate, this paper proposed a randomized approach to parity space-based fault detection. First, with no priori knowledge of the unknown inputs, a parity space-based residual generator was designed, and by applying Chernoff bound lemma, a randomized approach to the false alarm of detection was proposed. Then, to overcome the conservation of zero false alarm rate threshold, a threshold selection method with the allowance of certain fault false alarm rate was given to balance the tradeoff between false alarm rate and fault detection rate. Finally, a simulation experiment was performed with an unmanned aerial vehicle control system and the effectiveness of the proposed method was illustrated.